【摘要】 针对工业现场背景噪声强、波动大,且检测时间受生产节拍限制,机械运转噪声在线检测及故障诊断困难的问题,提出基于背景噪声修正法进行机械运转噪声在线测量,并通过对噪声信号的小波包分解,生成时—频相平面,将其转化为定量描述的布尔型矩阵,再对样本不同时频段进行加权处理后训练形成故障模板。将被测对象的噪声信号与模板进行比较,实现机械故障的诊断。该方法在我们研制开发的主减速器性能试验台上取得了良好的使用效果。更多还原

【Abstract】 Because of the difficulty to measure the noise of a running machine and diagnose its fault amid strong background noises, a method was proposed to correct the machine’s sound level according to the background noise, apply wavelet package decomposition to the noise signal, generate the Time-Frequency phase plane, allocate different weight factors to the different frequency bands according to its sensitivity to all kinds of faults and quantize it a boolean matrix. The samples are trained to be fault templates. The measured data were compared with the templates, which realized the fault diagnosis of machines along the production line. The use in the main reducer performance test bed developed by us shows that the method is effective and efficient.更多还原